Detection of Phishing Attacks Based on Improved Convolution Neural Networks with Attention Mechanism

Phishing attacks, a significant threat to cybersecurity, have been rapidly increasing in recent years. To mitigate this risk, more effective phishing detection techniques are urgently needed. However, current defense technologies remain inadequate. This paper proposes a novel approach that combines an attention mechanism with modified residual blocks within a conventional residual neural network. This modification addresses the vanishing gradient problem, enhancing the network's performance. Experimental results demonstrate that the improved model achieves superior classification accuracy compared to other state-of-the-art models.